STUDIES ON POLYPLOIDY I. CYTOLOGICAL INVESTIGATIONS ON TRIPLOIDY IN CREPIS M. NAVASHIN University op California Publications in Agricultural Sciences Volume 2, No. 14, pp. 377-400, plates 56, 57 Issued April 6, 1929 University of California Press Berkeley, California Cambridge University Press London, England STUDIES ON POLYPLOIDY I. CYTOLOGICAL INVESTIGATIONS ON TRIPLOIDY IN CREPIS BY M. NA YASHIN CONTENTS PAGE Introduction 377 Acknowledgments 378 Occurrence, morphology, and cytology of triploids 378 Cytogenetic behavior of triploids 380 Discussion and conclusions 383 Summary 393 Literature cited 395 INTRODUCTION Among chromosomal variations triploidy deserves first attention for several reasons. In the majority of cases it becomes a point of departure for further alterations. Except for tetraploidy, it repre- sents the most widespread type of chromosomal variation, at least in Crepis. The majority of other chromosomal aberrations occur in the progeny of triploid individuals and are therefore only conse- quences of triploidy. Moreover, being unbalanced, a triploid indi- vidual is incapable of sexual reproduction without segregation, and this unbalance is the very source of new chromosomal combinations in subsequent generations, some of which cannot originate in any other way. Among these derivatives from triploids should be mentioned the higher grades of polyploidy which are to be observed in the progeny of a triploid individual, as will be shown later. Some of them may be balanced and constant, the others in their turn may become sources of further chromosomal variation. These various chromosomal variations, including the higher grades of polyploidy, produce profound changes in the biological properties of the individual, affecting, as will be seen, the viability and the tempo of development, the conditions of pollination, the relations as to inter- specific crossability, etc. The influence of chromosomal variation on the size of organs and on fertility is well known. 378 University of California Publications in Agricultural Sciences [Vol. 2 All these consequences of triploidy, whether direct or indirect, make it an important phenomenon and one that should be kept in mind not only in theoretical work but also in practical breeding. It is sufficient here to mention the important services of trisomic and polyploid ratios in the task of localizing genes, the important results of investigations on genie balance and sex determination, etc. The value of seedless fruits and the superiority of triploid flowering bulbous plants give good illustrations of the practical importance of triploidy. Chromosomal variation also probably accounts for the origin of many new ornamental and commercial varieties. This paper contains the first report on investigations carried on during the period 1926-1928. The investigations were conducted mainly in the laboratories and experimental grounds of the Timiria- sev Federal Institute of Scientific Research and of the Comakademy in Moscow ; and were completed at the Division of Genetics, Univer- sity of California, under a fellowship of the International Education Board, New York. Acknowledgments During the course of the work the writer had the excellent assist- ance of Miss Gerassimova of Moscow, which it is a pleasure to acknowledge here. The data obtained by Miss Gerassimova will be mentioned in the course of this paper. The writer is especially glad to express his gratitude to Professor E. B. Babcock and Professor R. E. Clausen, of the University of California, for many valuable suggestions and for active interest in the work. OCCURRENCE, MORPHOLOGY, AND CYTOLOGY OF TRIPLOIDS In a study of extensive material of Crepis several triploid plants, Crepis capillaris (L.) Wallr., C. dioscoridis L., and C. tectorum L., were found, as well as other chromosomal variations. A preliminary note on polyploidy in Crepis (Navashin, 1925) and general observa- tions regarding the cytology and morphology of polyploids have already been published (Navashin, 1926). It has been shown that in populations of C. capillaris and C. tectorum triploid individuals are frequent, being sometimes present to the extent of one per cent of the total number of plants. Such a high degree of occurrence can- not be without some- influence on the genetic behavior of the species involved. 1929] NavasMn: Triploidy in Crepis 379 Crepis capillaris is especially favorable material for study, for it is distinguished from other plants which have been investigated hitherto by its low chromosome number (n = 3) as well as by very clear morphological features of the individual chromosomes, the latter circumstance permitting unmistakable identification even under most unfavorable conditions. This species has therefore been chosen as the principal subject of the investigations ; and triploidy has been investi- gated less fully in two other species, chiefly for the sake of com- parison with the conditions found in C. capillaris. As will be seen later such a comparison has proved interesting. A single triploid plant, "1947," of C. capillaris was the source of the material of the present investigation ; similarly a single plant of each of the other two species (C. tectorum and C. dioscoridis) ? The somatic chromosomes of triploid plants of C ' . capillaris are shown in plate 56. As may be seen from the drawings, the homo- logous chromosomes in a triploid individual are present in threes instead of twos; their size and shape as well as the most minute details of their organization (the satellites and their size) are wholly un- affected under triploid conditions. The metamorphoses of the chromo- somes during the prophase of mitosis differ in no way from those known to occur in a normal diploid nucleus. Special attention should be drawn here to the remarkable relation between the satellite and the nucleolus in the prophase of the somatic division. As was first pointed out sixteen years ago by S. Navashin (1912, 1927), the satellite originally appears in the prophase on the surface of the nucleolus, and becomes attached at a certain stage of karyokinesis by a very thin thread to a particular end of a specific chromosome. The writer's observations on triploids have revealed the same phenomenon with the sole difference that three satellites are formed instead of two as in diploids. Two stages of this process of satellite formation are shown in plate 56, a and b. It should be added, however, that such observations are rather difficult, success depending - entirely upon perfection of technique. The fixing fluid introduced by S. Navashin (chromic acid, formalin, and acetic acid in varied proportions), which has become popular among cytologists under various names, gives the best results inasmuch as the nucleolus becomes completely destained ; on the other hand, with Flemming's solution and most of the other usual fixatives the nucleolus is as deeply stained by haematoxylin as the chromosomes themselves. It 1 The first triploid plant of C. dioscoridis was found by Mrs. G. B. Medvedeva. 380 University of California Publications in Agricultural Sciences [Vol. 2 seems probable that the recent statements of Darlington (1926) deny- ing the formation of the satellites on the nucleolus are merely due to unsuitable or imperfect fixation. The somatic anaphase in a triploid plant also proceeds in a per- fectly typical way (pi. 56/). Consequently, the whole mitotic process in triploids appears to be normal even to the smallest details ; and the same is of course true also for the other two species, C. tectorum and C. dioscoridis. The triploid plants, like all other known triploids, are distinguished by notably increased dimensions of the cells and cell organs. In these triploids the fruits and other external features are also enlarged ; in fact the entire plant is somewhat larger than normal. The fertility is greatly reduced and a high percentage of the pollen is bad ; all these features are well known and should be considered as characteristic consequences of triploidy. Pertinent data are pre- sented in table 1. TABLE 1 The Sizes of Triploid C. capillaris Plants, the Weight of their Achenes and Percentage of Good Seeds, as Compared with the Normal Diploid Sister Plants of the Same Species Data of Miss Gera.ssim.ova. Triploids Diploids (average of 33) (average of 11) Height of the plants 88.2 cm. 69.7 cm. Weight of 100 achenes 0.0439 gr. 0.0354 gr. Percentage of good seeds 21.8 65.8 Besides the enlarged dimensions and reduced fertility the triploid plants were distinguished by their slow but robust growth. The majority of the triploids started to bloom about a month later than the diploid sister plants, although planted at the same time and grown under exactly the same conditions. CYTOGENETIC BEHAVIOR OF TRIPLOIDS As has been shown above, tbe triploid plant in spite of its reduced fertility sets enough seeds to produce numerous progeny. All the seeds (149) secured from open pollination of the original plant, of C. capillaris were planted and produced 107 plants, two of which died. The chromosomal constitution of the remaining 105 plants is presented in table 2. 1929] Navashin: Triploidy in Crepis 381 TABLE 2 The Chromosomal, Constitution of the F, Plants Produced by Open Pollination of Triploid C. oapillaris Plant, ' ' 1947 ' ' Chromosomal constitution 2n 2n + I - 01 Number 70 2 served Per cent 66.7 1.9 31.4 Expected dist Number 13.125 39.375 39.375 13.125 from ribut binomial ion Per cent 12.5 37.5 2n + I + I 37.5 3n 33 105 12.5 Total 100.0 105.000 100.0 The F 1 plants were further investigated. From each of them a sample of 100 open-pollinated seeds was taken and the root tips secured after planting were studied cytologically. The chromosome constitution of the F. 2 is tabulated in table 3. The figures of this table, as may easily be seen, are essentially equivalent to those pre- sented in table 2. As with F t , the F„ consists of a great majority of diploids and triploids (888 diploids and triploids out of the total 959 plants) and only a negligible number (29 out of 959, i.e., about 3 per cent) of simple and double trisomies. It differs from Fj only in the presence of a certain number of tetrasomics, tetraploids, and a 'single 7n plant. In order to get data more nearly complete the triploid plants were crossed by Miss Gerassimova inter so and also with diploid sister plants. Many trisomic and polyploid plants were obtained from these crosses and were carefully studied. The data concerning these experi- ments will be reported in a later paper. TABLE 3 Chromosomal Constitution of E 2 Plants ; cf. Table 2 Data of Miss Gerassimova. ( 'tiromosomal constitution 2n 2n + I 2 n + I + I 3n 3n + T 3n + I + I 4n 7n Observed umber Per cent 613 63.9 13 1.3 16 1.7 275 28.8 1 0.1 2 0.2 38 3.9 1 0.1 Expected from binomial distribution Number Per cent 119.875 12.5 359.625 37.5 359.625 37.5 119.875 12.5 Total 959 100.0 959.000 100.0 Of the six theoretically possible trisomic types in C. capillaris five were actually obtained and grown under controlled conditions. These 382 University of California Publications in Agricultural Sciences [Vol. 2 types were as follows (the capital letters, A, C, and D, representing the three chromosome types) : simple trisomies, triplo-A and triplo-Z) and double trisomies, triplo-AC, triplo-AZ) and triplo-CZ). The ex- pected triplo-C simple trisomic type has not been obtained, although each of the remaining five types was found repeatedly. Plate 57 illustrates the somatic chromosomes of the trisomic types, together with the normal diploid chromosome complex. The trisomies are easily distinguished from the normal and triploid plants. Besides their lower viability and slower growth, they differ from diploids and triploids in the shape and color of the leaves and in a number of other features. Different chromosomal types of trisomies differ strikingly in their morphology, and can be readily recognized without cytological investigation. Some of them approach the triploid in fertility. A complete report of the investigations of these trisomic types will be given in a later paper. The triploid plants of C. capillaris were finally crossed by Miss Gerassimova with normal plants of various other Crepis species; pri- marily in the hope of determining the constitution of their gametes, and also in order to test their crossability, which might differ from that of normal plants. The majority of interspecific hybrids obtained from the application of the foreign pollen to the stigmas of the the triploids proved upon examination to contain a diploid chromo- some complex of C. capillaris together with a haploid chromosome group of the other species used as the male parent ; moreover, viable hybrids were obtained from crosses which had never been successful when diploid C capillaris plants were used for crossing. Thus a hybrid between C. capillaris and C. alpina was obtained, a cross which had never been secured before. It has thus been shown that a triploid plant undergoes interspecific hybridization more readily than a diploid one. Finally, a hybrid plant was obtained which possessed four haploid chromosome complexes of C. capillaris and one of another species (C neglecta). This demonstrates that triploids may produce viable tetraploid eggs in addition to haploid and diploid ones. These polyploid hybrids of various kinds proved to be much more fertile than the common diploid ones. For comparison with the conditions discovered in C. capillaris the progenies of triploid plants of C. tectorum and C. dioscoridis were studied. The behavior of these species proved to be entirely different from C. capillaris. The majority of the plants in the progeny of triploid C. tectorum consisted of various trisomic types, and only a 1929] Navashin: Triploidy in Crepis 383 relatively small number of diploids and triploids occurred. In C. dioscoridis the diploid plants predominate, and trisomies occur only in small numbers, approximately equal to the number of triploid plants. In addition to trisomies and other variations resulting from numeric changes of whole chromosomes among the progeny of triploid plants, cases were found of alterations in the chromosomes them- selves. Thus in C. capillaris one plant was found to possess a frag- mented D-chromosome. This particular plant possessed a very small satellited chromosome (<7), representing the proximal part of the normal satellited chromosome, which behaved as a new autonomous chromosome. Similar fragmentation of the satellited chromosome has been found to occur in the progeny of triploids in C. feci ovum. In the latter species several other alterations in chromosome organiza- tion have also been observed (1926). One case of fragmentation is illustrated in plate 57/. Finally, several spontaneous interspecific hybrids have been found in the progeny of triploid plants. The majority of them possessed two haploid complexes of one species and one of the other ; but a few were also found which possessed three haploid sets of one species and one of the other. DISCUSSION AND CONCLUSIONS The manner of origin of the original triploid C. capillaris plant ("1947") could not be finally established. Since triploidy, in con- trast to tetraploidy, cannot be a consequence of a purely vegetative process, the "summation" of three haploid chromosome complexes in a zygote should theoretically be due to one of the following causes : viz., (1) formation of a diploid gamete followed by fusion with a normal haploid one; (2) dispermy; and (3) formation of an embryo from a cell of the endosperm. In all the triploid individuals known to have arisen under experi- mental conditions the manner of origin has been proved only in the cases of triploid Oenotheras and Daturas, the plants being obtained artificially from crosses of tetraploids with diploids (Geerths, 1911; Blakeslee, 1924). There is no doubt, consequently, that in these particular instances triploidy is due to fusion of a diploid gamete with a haploid one. As for the other well-known instance of triploidy 384 University of California Publications in Agricultural Sciences [Vol.2 found to occur spontaneously in cultures of Oenothera Lamarkiana, its origin still remains somewhat obscure. Based on the observations of B. Nemec (1910) and Ishikawa (1918) on Gagea and Oenothera, R. Gates (1924) suggested that dispermy should be responsible for triploidy. The majority, however, are inclined to accept the same manner of origin as in experimental triploids, viz., occurrence of diploid gametes. The third conceivable explanation, viz., formation of the embryo from a cell of the endosperm, remains purely specu- lative since such a phenomenon has never been observed in any plant. In the case of Crepis the writer does not hesitate to suggest the first method, for several spontaneous "triploid" hybrids have been found which have arisen through the open pollination of diploid plants of C. capillaris (Navashin, 1927). It is obvious that such interspecific hybrids could not obtain a diploid capillaris complex except from a diploid egg cell; it is quite improbable that two sperms, one belonging to C. capillaris and the other to another species, could fertilize a normal egg, and thus give rise to a triploid hybrid. The occurrence of occasional functional diploid egg cells being clear enough, the question arises as to the influence of the increased chromosome number on the sexual function of the female gamete. R. Gates (he. cit.) on the basis of the data on apogamous plants doubts the very possibility of fertilization of a diploid egg, the Latter circumstance beinp', he thinks, an indirect proof of his hypothesis of dispermy. It has been shown, however, that diploid eggs are capable of fertilization as well as haploid ones; and moreover, it has been possible to demonstrate that not only diploidy bul even higher grades of polyploidy do not affect the sexuality of the gamete. Pollinating triploid capillaris (after the usual castration) with the pollen of other Crepis species usually gave triploid hybrids, but, as shown above, a few plants possessing more than two haploid capillaris chromosome complexes were obtained. As a result of crossing two triploid plants together, not only triploid and tetraploid but also pentaploid plants were obtained. Finally, as has been shown in table 3, a heptaploid (7n) plant of C. capillaris was found in the immediate progeny of a triploid. Prom these results it is evident that besides diploid egg cells triploid, tetraploid, and possibly even pentaploid or hexaploid ones are formed, and that these polyploid egg cells undergo normal fertil- ization, producing zygotes possessing as highly multiplied chromo- some complexes as 7n, i.e., twenty-one chromosomes instead of the 1929] Navashin: Triploidy in Crepis 385 normal diploid number, six. How far increase in chromosome material may go in Crepis is still unknown. These curious results clearly demonstrate once more, that increase of chromatin material of itself cannot produce development of the zygote because the diploid, triploid, tetraploid, and perhaps pentaploid or hexaploid eggs are normally incapable of development without fer- tilization. Apparently the male gamete contributes some stimulating materials besides the usual chromatin materials. Thus direct cyto- logical study supports once more Loeb's famous conclusions. We may conclude, therefore, that triploidy in Crepis is due to the occasional formation of diploid gametes, probably of the female ones. There is a possibility, however, that an occasional diploid pollen grain may function, although it has not been certainly demonstrated. As to the manner of origin of diploid gametes three possibilities may be suggested, viz.: (1) omission of the reduction division; (2) duplication of the reduced (haploid) nucleus; and (3) formation of tetraploid groups of somatic cells followed by normal reduction during sporogenesis in the resulting tetraploid tissues. The data presented here being insufficient to arrive at a definite conclusion, one may equally suggest any one of these three conceivable ways, inasmuch as all of them are known to take place. Omission of the reduction division, particularly under the influence of various chemical and physical factors, is a well-known phenomenon ; forma- tion of groups of tetraploid cells, besides other numerous instances, has also been found in Crepis (M. Navashin, 1926; Hollingshead, 1928a) ; and duplication of the haploid nucleus in the female gameto- phyte has been recently confirmed by Newton (1927). It is interesting to point out that cytology provides a unique method of demonstrating the mode of origin of a diploid gamete when other methods fail or for some reason are inapplicable. If the homo- logous chromosomes in the same nucleus possess distinguishing features one may directly arrive at a definite conclusion as to the derivation of the diploidy of the gametes, and consequently of the triploidy. Advantage may be taken of the occurrence of size differ- ences of the satellites of homologous chromosomes, such as have been demonstrated above. If the original plant possesses unequal satellites the occasional diploid gametes produced by it will be different accord- ing to their mode of origin. Thus if diploid gametes are produced by non-reduction they will uniformly contain a pair of unequal satel- lites. On the other hand, if diploid gametes are formed as a result 386 University of California Publications in Agricultural Sciences [Vol. 2 of somatic duplication before reduction they will be of three different kinds: viz., (1) both satellites large; (2) both satellites small: and (3) unequal satellites. Finally, if duplication of a reduced nucleus has taken place, the gametes will be of two different kinds either with both satellites small or with both satellites large. If the original plant with unequal satellites is crossed with a plant having equal satellites one can easily get direct evidence as to the manner of origin of the occasional triploids. If among the triploids there are plants with three equal satellites, the cause of triploidy cannot be due to non-reduction, but if there are no plants of that sort then triploidy must be due to non-reduction. Taking special precautions one can even find out whether duplication took place before or after reduc- tion. Special experiments in this direction are in progress. No matter what the manner of origin of triploidy may be, it may undoubtedly play a considerable role in the genetic behavior of the species involved. Besides the well-known peculiarities of Mendelian inheritance caused by new trisomic and polyploid chromosome com- binations, attention should be drawn here to some other consequences of triploidy. In table 3 (p. 381) it is shown that 92.7 per cent of the entire progeny of triploid plants of C. capillaris consists of normal diploid plants (63.9 per cent) and triploid plants (28.8 per cent) ; and the remaining 7.3 per cent of higher grades of polyploidy (4.0 per cent), various trisomies (3.0 per cent), and triploid tetrasomics (0.3 per cent). Prom the same table it may be seen that these figures are not in agreement with those expected from the binomial distribution of the extra haploid complex (see table 3, column "expected"), but are in conflict with it in several respects. Thus instead of the expected 75 per cent of trisomies only 3 per cent were actually obtained; and instead of 25 per cent of diploids and triploids, as a matter of fact there are 92.7 per cent. In order to explain the deficiency of trisomies the percentage of aborted seeds was determined under the suppo- sition that they should represent elimination of eggs bearing 1 or 2 extra chromosomes. As may be seen from table 1 the triploid plants set on the average only 21.8 per cent of good seeds; the lack of 72 per cent of the expected trisomies is therefore in full agreement with the observed reduction in the number of good seeds. It is further very probable that the lack of trisomies may be due to partial zygotic sterility arising from disturbances which in the majority of instances 11)29] Navashin: Triploidy in Crepis 387 prevent the very formation of gametes bearing extra chromosomes; for the rare instances in which gametes are formed result in perfectly viable zygotes; simple or double trisomic individuals. On the other hand, there is a possibility of selective fertilization, although the high percentage of bad pollen in triploids makes it improbable that many pollen grains possessing extra chromosomes could exist. Further difficulties are met with in explaining the observed numeric ratio of diploids and triploids which, as may be seen from table 3, is almost exactly equal to 2:1, instead of the expected 1:1 ratio. Although the observed ratio is similar to the usual zygotic lethal ratio, there is no reason to suggest any lethal factor of that sort, for under triploid conditions it could not give rise to the expected 1 : 2 ratio. A gametic lethal factor of some sort could, of course, give a 1 : 2 ratio, but the first following generation would immediately revert to the normal 1 : 1 ratio, on account of the loss of the lethal. Similarly it is impossible to suggest any stimulating factors which would make the haploid cell win in the struggle for life in the female tetrad in the young ovule. The elimination of trisomies should not, of course, affect the ratio of diploid and triploid offspring. Although it is impossible, from the purely theoretical point of view, to account for this curious ratio which occurs quite regularly in F t and F 2 (tables 2 and 3), two reasonable hypotheses may be suggested: 1. If the odd members of the triploid complex of C. capillaris are regularly distributed to the poles, the extreme variants, i.e., the haploid and diploid germ cells, will be formed in equal numbers. On the other hand, if the odd chromosomes sometimes lag, one of the sister cells of the dyad will receive an incomplete set of chromosomes instead of a full diploid complex. Cells of that sort, however, are almost incapable of further development as is evident from the absence of trisomies. It is possible that such a female dyad will give rise to a haploid embryo sac, no matter what the position of the haploid component may be, i.e., whether the haploid cell be turned toward the microphile or toward the chalaza. Such a phenom- enon will cause an increase of the number of haploid eggs, and con- sequently of the resulting diploid plants. Similarly, lagging of chromosomes in the intermediate types of distribution may lead to an increase of the relative number of haploid gametes. Despite the reasonableness of such an explanation it still remains quite obscure why such disturbances in the reduction division should cause the observed regular ratio. 388 University of California Publications in Agricultural Sciences [Vol.2 2. If one suggests the existence of a factor a double dosage of which stimulates parthenogenetic development of a diploid egg, it is very easy to explain the observed 2 : 1 ratio. Assuming that two homologous chromosomes of the original triploid plant contain this factor, the gametic ratio among diploid eggs will be as follows: 2 eggs bearing a simple dosage of the factor : 1 bearing a double dosage. Since the latter should develop into diploid plants without fertiliza- tion, the proportion of diploids will increase one-third at the expense of the triploids; as a consequence the ratio of diploids and triploids will be equal to 2 : 1, instead of 1: 1. If, however, such an explana- tion is correct, different ratios should be expected among progenies of different triploid plants, according to their genetic constitution ; plants possessing a double dosage of the "parthenogenetic" factor should give the 2:1 ratio, while those possessing a single dosage should produce diploid and triploid offspring in equal numbers. Among the 34 plants investigated 32 have not shown a 1:1 ratio but, on the contrary, the progeny of these 32 plants approximated more or less closely a 2:1 ratio; only two plants produced diploid and triploid offspring in equal numbers, the latter circumstance being perhaps merely accidental. The reduction division being not yet completely investigated, it is impossible to say whether or not lagging of chromosomes may account for the observed ratio. On the other hand, some experimental evidence makes it probable that parthenogenetic development of diploid eggs does take place. Furthermore, it has been found that the triploid plants set a high number of good seeds if attempts are made to cross them to certain other Crepis species; instead of hybrids, however, these seeds produce only normal diploid C. capillaris plants. This phenomenon was especially striking when C. rubra was used as the pollen parent. The production of pure capillaris plants may be explained, of course, simply as an experimental error due to incom- plete depollination ; on the other hand, however, it seems rather strange that the progeny obtained in this way should consist only of diploid plants, while it is known that under normal conditions about one-third of the progeny always consists of triploids. It seems to be probable, therefore, that parthenogenetic development occurs possibly under the stimulating influence of the foreign pollen, which is, how- ever, incapable of effecting fertilization. Jorgensen's observations (1928) on the parthenogenetic formation of diploid offspring in Solatium give a good illustration of the actual occurrence of the 1929] Navashin: Triploidy in Crepis 389 process discussed above. This suggestion becomes still more probable if one reoalls the occurrence of haploidy in Crepis recently discovered by Miss Hollingshead (19286), which removes all doubt as to the possibility of parthenogenetic development of the egg in Crepis. Essentially different conditions have been disclosed by Belling and Blakeslee (1922) in triploid Daturas. The majority of the progeny of triploids consists of simple and double trisomies together with diploids, the latter being present only in the amount of about one-third of the total number of plants, and triploid plants do not occur at all. As for the other expected combinations as well as the higher grades of polyploidy (excepting tetraploidy), they were not observed in the triploid progenies in Datura at all. In triploid Oenothera (Lamarkiana var. semiyigas) according to van Overeem (1920) the formation of all possible chromosomal com- binations has been observed in ratios approximating those expected from the binomial distribution. In triploid tomatoes according to J. W. Lesley (1928) only single, double, and triple trisomies occur, altogether amounting to about 85 per cent of the progeny ; the remaining 15 per cent consisting of normal diploid plants. No tetraploids or higher grades of polyploidy were observed. In other Crepis species, as shown above, the conditions are differ- ent from those existing in C. capiltaris. In C. tectorum triploids behave like those of Oenothera; in C. dioscoridis they produce few trisomies and few triploids, the great majority of the offspring being normal diploid plants. It is clear, therefore, that triploid representatives of different species, even though closely related, may differ strikingly in the types of progeny which they produce. As respects the biological importance of triploidy, attention should be drawn first to the manifold instances of chromosomal variation which are known to occur in the progeny of triploid plants. As has been shown above, in addition to tetraploids, which are known to occur in a number of other species, pentaploid and even heptaploid plants are produced in the progeny of triploid plants of C. capiUaris. Triploidy represents, therefore, the initial step of the accumulation of the chromatin material ; how far this process can go is not yet known, but there is some evidence suggesting the possibility of far higher grades of polyploidy than those which have been found to date. Moreover, the tendency of triploids to produce polyploid interspecific 390 University of California Publications in Agricultural Sciences [Vol. 2 hybrids makes it probable that triploidy should be accounted one of the initial steps in the development of such Crepis species as C. biennis and C. ciliata both possessing about 20 pairs of chromosomes. As for the manner in which a polyploid race or a polyploid individual may become an initiator of a new species, i.e., whether mutation, hybridization, or both together may produce the specific differences involved, that is still obscure. It may be considered as proved, however, that polyploidy in Crepis causes profound changes in the biological properties of the species involved. It should be emphasized here that a slowing down of the rate of ontogenetic development occurs as a consequence of polyploidy. It is perfectly clear that a delay of anthesis of only two weeks may under certain conditions give a considerable advantage to the plant. The rate of development may be affected even far more by polyploidy, a circumstance which may play a decided role in the invasion of an area by a new polyploid form, if this area were formerly unavailable on account of climatic conditions. If the new- form is able to conquer a new territory, it will probably undergo there the influence of a series of new conditions which were absent in the original area. A result of such extension of the geographic area occupied by a given species, which might be of the first impor- tance in the evolution of a genus, would be the meeting and conse- quent hybridization of species which have not previously been in con- tact. If the new polyploid form remains in the original area, it will nevertheless meet new opportunities of crossing with other later blooming species, due to the delay of anthesis. The foregoing sug- gestions apply to all weeds controlled by periods of grass cutting or time of harvest as well as to other plants independent of the agri- cultural activity of man. The importance, to agriculture, of the rate of development of commercial plants is of course well known. It is of interest here to point out that the climatic frontier pre- venting the further advance of a given form may itself become a place of origin of polyploids due to peculiarities of temperature influ- ences; for it is very well known from experimental investigations that unusual temperature conditions may be an important cause of the production of polyploid gametes (de JInl, 1923; Belling, 1925). After migi-ation the polyploid forms thus produced on the limits of distribution will be isolated, the climatic barrier preventing them from meeting with the original form. The new additional polyploids eventually formed along the limits of distribution might steadily 1929] Navashin: Triploidy in Crepis 391 penetrate into the barrier zone, thus continuing the process of in- vasion. Some evidence confirming these suggestions may be seen in C. biennis. This polyploid species (J. Collins and M. Mann, 1923), owing perhaps to its biennial life-cycle, has acquired the ability to inhabit an area which is unavailable to another closely related species, C. capillaris. Furthermore, as was shown above, triploidy stimulates interspecific crossability, probably by reason of the higher viability of the hybrid zygote derived from a diploid egg, and the presence of a normal balanced chromosomal complex of one of the species crossed. Other features may also play a considerable part ; among them the partial male sterility of triploids, which may facilitate successful cross- pollination. There is a possibility, moreover, that the structure of the stigma and of the style facilitates hybridization between some species, owing to enlarged dimensions of the cells, etc. Such polyploid hybrids, as has been shown, are not only more viable, but, moreover, show a higher degree of fertility than normal ones; and the peculiarities of their chromosomal constitution afford favorable conditions for the production in subsequent generations of various recombinations of chromosomes of the parental species. Finally, crossing of two triploid individuals belonging to different species may produce directly a balanced amphidiploid (M. Navashin, 1927) hybrid, as a result of the meeting of two diploid gametes. The results of investigations on polyploid interspecific hybrids in Crepis will be reported in another paper. The other combinatory alterations of the cell nucleus occurring in the progeny of triploids, i.e., trisomies and tetrasomics of various kinds, could hardly play any part in species formation. For Datura there is even direct evidence against such a suggestion, for balanced tetrasomic types show greatly reduced viability, and it is unreasonable to ascribe to them the role of originators of new forms. In Crepis, as a matter of fact, such types do not even occur. For the purpose of genetic analysis the trisomies in Crepis are of the highest interest because of the fact that the individual chromo- somes may be easily identified. It should be pointed out that the excellent morphological features of the chromosomes in Crepis allow a visual verification of the hypothesis of the origin of secondaries, the latter being supposed according to Belling (1924) to have duplicated ends of certain chromosomes. A careful study of trisomies in Crepis shows, however, that none of them possess even the slightest abnor- 392 University of California Publications in Agricultural Sciences [Vol. 2 malities in chromosome organization, not to mention duplication of the ends of the chromosomes. One must conclude, therefore, that either there are no duplicational secondaries in Crepis or their nature should be explained in some other way. It is hoped that further genetic study on trisomies will throw light on this question. The frequencies of different trisomic types is not the same. In C. capillaris apparently the triplo-D type (possessing one satellited chromosome extra) is the most common one, while the triplo-A type occurs less frequently. Finally, the third simple trisomic type, triplo- C, probably occurs extremely rarely, if at all ; at least, it was not found although numerous trisomies of the other two types were dis- covered. Consequently, C. capillaris exhibits some analogy in this respect to Drosophila melanogaster; for according to Bridges (1923) individuals with extra II or III chromosomes do not occur in the progeny of triploid females but triplo-X and triplo-IV flies are fairly common. The lack of these combinations in Drosophila is reasonably explained by the disturbances produced by the unbalanced excess of large chromosomes ; in the case of Crepis, however, such an expla- nation cannot be maintained, for here the individuals possessing the smallest chromosome in excess are lacking. Evidently the reason for the absence of certain chromosomal types does not depend entirely upon the relative amount of chromatin present. Finally, as was shown above, alterations occur in the organization of the chromosomes themselves in the progeny of triploids. The most interesting variation of that sort is the fragmentation of the satellited chromosome, the latter phenomenon leading to a formation of one very short satellited chromosome (ef. pi. 571) and another longer one, instead of the original single long satellited chromosome. This par- ticular alteration has been found to be the most common one among other instances of reorganization of chromosomes and has been dis- covered in two species (C. capillaris and C. tectorum). The relative dimensions of both fragments are different in different individuals; consequently the writer's original suggestion (1926) that there should be a certain "point" in the chromosome where the fragmentation takes place, must be considered incorrect. The cause of this frag- mentation is still obscure ; perhaps it is in some way connected with the peculiar behavior of the satellited chromosomes which, as was shown above, are represented in the earlier stages of division by two disconnected parts (the satellite and the body of the chromosome itself) and which are also subject to striking changes in hybrid nuclei 1929] Navashin: Triploidy in Crepis .°>93 (M. Navashin, 1927). At any rate, there is no doubt that the unbalanced conditions of the nucleus in some way stimulate chromo- some alteration. Although the viability of individuals possessing fragmented chromosomes is greatly reduced, there is still some prob- ability that fragmentation of some sort may take part in species formation. In this connection one may refer especially to C. parvi- flora, which possesses an extremely short satellited chromosome and another longer one, both suggesting in a way the products of frag- mentation of the usual satellited chromosomes of C. capillaris or C. tectorum. It is of interest to note that the same process of frag- menation has also been reported in progenies of triploid tomatoes by J. W. Lesley (he. cit.). SUMMARY 1. Triploid individuals have been found in many populations of Crepis capillaris, C. tectorum, and C. clioscoridis, in some instances in proportions up to one per cent of the total number of plants. Such a high degree of occurrence cannot fail to influence the biological behavior of the species involved. 2. Triploid individuals differ morphologically from normal ones only in epiantitative features ; viz., in enlarged general dimensions and organs, in enlarged cells and cell organs, in increased size of fruits, etc. 3. The tempo of development of the triploid plants is significantly delayed, although the viability is not perceptibly reduced. 4. The fertility of triploids is greatly reduced ; triploid plants of C. capillaris produce on the average only about 21.8 per cent of good seeds, because of failure of the majority of ovules to function. Simi- larly the majority of pollen grains are aborted. 5. Cytological investigations of triploids has shown that their chromosomes do not differ in any way from those of diploids except they are present in triple number. 6. The first and second generations obtained from the original triploid C. capillaris plant have been investigated cytologically, the total number of plants examined exceeding 1000. In C. capillaris 92.7 per cent of all the progeny of triploids consists of diploids (63.9 per cent) and triploids (28.8 per cent) ; the remaining 7.3 per cent 394 University of California Publications in Agricultural Sciences [Vol.2 comprises simple and double trisomies (3.0 per cent), triploid tetra- somics (0.3 per cent), and higher grades of polyploidy (4.0 per cent). The cause of these curious ratios is still unknown. 7. Of the six possible trisomic types in G. capillaris five have actually been found and grown under controlled conditions. These different trisomies differ morphologically from one another; some of them are usually not less fertile than the triploid plants. 8. The other two species (C. tectorum and C. dioscoridis) behave in quite a different way. The progeny of triploid tectorum plants consists of a majority of trisomies, the diploid and triploid plants being present in the minority. In C. dioscoridis, on the other hand, the progeny of triploids contains only a small percentage of trisomies and triploids; the great majority of the offspring arc diploid. 9. Crossing of triploid plants with other Crepis species has shown that the triploid condition in the female parent is favorable for inter- specific hybridization. From these crosses certain hybrids have been obtained which could never be obtained heretofore by using normal diploid plants. Many of the hybrids thus secured were polyploid. These properties of triploids under natural conditions may play ;i considerable part in species formation; they also make triploids favorable material for theoretical genetic work as well as for prac- t teal breeding. 10. The formation of diploid gametes (most likely of female ones) should be considered as the original source of triploidy in Crepis. 11. In a number of instances the occurrence of triploid, tetraploid, and possibly pentaploid and hexaploid eggs has I n proved; these polyploid gametes are perfectly viable and arc capable of fertilization. 12. Triploidy in Crepis may play a considerable although indirect part in formation of new species, for it gives rise in subsequent gen- erations to a number of further chromosomal variations, especially of higher grades of polyploidy. Through change in the rate of devel- opment, a polyploid individual may acquire the ability of withstand- ing different climatic conditions. As a consequence it may penetrate into a new territory where it may become subjeel to a series of various influences, especially to crossing with other species and forms. On the other hand, the climatic barrier may be considered as a possible source of new polyploid forms arising in response to extreme tempera- ture conditions; and at the same time this barrier will provide perfect isolation. 1929] Navashin: Triploidy in Crepis 395 13. It is difficult to imagine how the resulting trisomic types can play any role in species formation, although they represent first-class material for genetic analysis, thanks to the clear morphological differ- ences of the chromosomes. There is some probability, however, that fragmentation of chromosomes may take part in the formation of certain Crepis species. LITERATURE CITED Belling, John 1925. The origin of chromosome mutants. Jour. Gen., 15:245—266. Belling, John, and Blakeslee, Albert F. 1922. The assortment of chromosomes in triploid Daturas. Am. Nat., 56: 339-346. 1924. The configuration and size of the chromosomes in trivalents of 25- cliromosome Daturas. Proc. Nat. Acad. Sci., 10:116-120. Blakeslee, Albert F., and Belling, John 1924. Chromosomal mutations in Jimson weed, Datura stramonium. Jour. Hered., 15:195-206. Bridges, Calvin B. 1921. Genetical and cytological proof of the non disjunction of the fourth chromosome of Drosophila melanogaster. Nat. Acad. Sci. (U.S.A.), Proc, 7:186-192. 1923. Aberrations in chromosome materials. Eugenics, Genetics and Family (Second Internal. Congress Eugenics), 1:76-77. Collins, J. L., and Mann, Margaret C. 1923. Interspecific hybrids in Crepis, II. Genetics, 8:212-232. Darlington, C. D. 1926. Chromosome studies in the Scilleae. Jour. Gen., 56:238—251. Gates, Ruggles E. 1924. Polyploidy. British Jour. Exper. Biol., 1:153-182. Geerths, T. H. 1911. Zvtologische Untersuclmngen einiger Bastarde von Oenothera gigas. Ber. deutsch. Botan. Ges., 29:160-166. Hollingshead, Lillian 1928a. Chromosomal chimeras in Crepis. Univ. Calif. Publ. AgT. Sid., 2: 343-354. 1928?). A preliminary note on the occurrence of haploids in Crepis. Am. Nat., 62:1-4. ISHIKAWA, M. 1918. Studies on the embryo sac and fertilization in Oenothera. Ann. Bot., 32:279-317. J0RGEXSEX, C. A. 1928. The experimental formation of heteroploid plants in the genus Solanum. Jour. Gen., 19:13:5-210. Lesley, J. W. 1928. A cytological and genetical study of progenies of triploid tomatoes. Genetics, 13:1-43. 396 University of California Publications in Agricultural Sciences [Vol. 2 Mol, W. E. DE 1923. Duplication of generative nuclei by means of physiological stimuli and its significance. Genetica, 5:225. Navashin (Nawaschin), M. 1925. Polyploid mutations in Crepis. Genetics, 10:583-592. 1926. Variabilitat des Zellkerns bei Crepis-Arten in Bezug auf die Artbildung. Zeitschr. Zellforsch. Mikr. Anat., 4:171-215. 1927a. tiber die Veranderung von Zahl und Form der Cliromosomen infolge . I < • r Hybridisation. Ibid., 6:195-233. Navashin (Nawaschin), S. 1912. On the nuclear dimorphism in somatic cells of Galtonia candicans (Eussian). Bull. Acad. Sci., 22. 1927t». Zellkerndimorphismus bei Galtonia candicans Des. und oinigen ver- wandten Monokotylen. Ber. deutsch. Bot. Ges., 45:415-428. Nemec, B. 1910. Das Problem der Bef rucht ungsvorgange, etc. (Berlin), 532 pp. Newton, W. C. F. 1927. Chromosome studies on Tulipa and some related genera. Linnean Soc. Jour., 47:339-354. Overeem, Caspar van 1920. t v ber Formeri mit abweichendea Chromosomenzahl bei Oenothera. Bot. Centralbl., Beihefte, 38:13-113. EXPLANATION OF PLATES PLATE 56 Somatic chromosomes of triploid plants of Crepis capillar is. X 3650. a, Prophase showing the satellites on the surface of the nucleolus. b, Later prophase showing the attachment of satellites to chromosomes. c, Metaphase showing the chromosomes of the original triploid plant "1947." Note the differences in sizes of the satellites. d, Chromosomes of a F, plant possessing a set of satellites identical to the original plant. e, Chromosomes of another F t plant possessing two small and one large satel- lites. /, Early anaphase showing the regular splitting of the chromosomes and of the satellites. |*».S] UNIV. CALIF PUBL. AGR. SCI. VOL. 2 NAVASHINl PLATE 56 d^ / PLATE 57 Somatic chromosomes of different derivatives produced by a triploid C. capillaris plant compared with a normal diploid chromosome complex (the capital letters representing the three chromosome types: A, the largest chromosome, C, the smallest chromosome, and D, the satellited chromosome). X 3650. g, Triplo-A (simple trisomic). h, Triplo-D (simple trisomic). i, Triplo-AC (double trisomic). j, Triplo-AD (double trisomic). k, Triplo-CD (double trisomic). I, Chromosomes of a triploid plant possessing an additional proximal fragment of the satellited chromosome. m, Chromosome complex of a normal diploid plant. [400] UNIV. CALIF. PUBL. AGR. SCI, VOL. 2 NAVASHINl PLATE 57 X w |C n y r\ ?/,